Avertissement sur les risques et la méthodologie

1. Nature of the output

Every finding is labelled by evidence class so you can judge how much to trust it:

• Deterministic findings — read directly from the market's own rule text, resolution source, metadata, or order book. Exact and reproducible (e.g. an exact rule fragment, a best bid/ask).
• Derived findings — computed from deterministic inputs (e.g. executable price at a given size, extracted deadlines, market relationships).
• Heuristic / modelled findings — involve interpretation or classification (e.g. the resolution-risk score, anomaly labels, Smart Flow inference). These are estimates, not facts.

2. The resolution-risk model

The composite dispute-risk score is a statistical model fitted on historically resolved markets. It currently beats a constant base-rate baseline out-of-sample (held-out Brier 0.119 vs 0.149, and beats the category, segment and rules-only baselines too), which is why it is labelled calibrated. Even so, it is a probability estimate over an uncertain future — not a guarantee that a market will or will not be disputed. If a future dataset revoked that verdict, we would relabel it MODEL_NOT_CALIBRATED and say so. Historical associations are shown with their sample size and confidence interval, and are associations, not proof of causation.

3. Prices & probabilities

Where the Service shows a market-implied probability, best bid, best ask, midpoint, or executable price, it is a read of public market data at the stated timestamp — not our own forecast and not a promise of execution. Depth, slippage, and partial-fill figures are modelled from the visible order book and may differ from a real fill.

4. Data freshness & coverage

Results may be live, cached, or historical, and coverage is partial — not every market has complete rules, a resolution source, or a live book. Each result states its evidence timestamp and, where relevant, that coverage is incomplete. See Coverage. We never fabricate an analysis when upstream data is unavailable; we say the evidence is insufficient.

5. Third-party data

Market facts come from Polymarket, UMA, and public blockchains via third-party providers we do not control. Their data can be delayed, incomplete, or wrong, and their terms and availability are their own.

6. Your responsibility

Prediction markets carry risk of total loss. Any decision you make is your own. Use the Service as one input to your own research, verify anything material against the primary source, and never rely on it as a guarantee. Methodology detail: Methodology.